From: Daniel Phillips <phillips@bonn-fries.net>
To: William Lee Irwin III <wli@holomorphy.com>, linux-kernel@vger.kernel.org
Subject: Re: hashed waitqueues
Date: Sat, 5 Jan 2002 03:44:06 +0100 [thread overview]
Message-ID: <E16Mgoj-0001Ew-00@starship.berlin> (raw)
In-Reply-To: <20020104094049.A10326@holomorphy.com> <E16MeqE-0001Ea-00@starship.berlin> <20020104173923.B10391@holomorphy.com>
In-Reply-To: <20020104173923.B10391@holomorphy.com>
On January 5, 2002 02:39 am, William Lee Irwin III wrote:
> 2 or 3 shift/adds is really not possible, the population counts of the
> primes in those ranges tends to be high, much to my chagrin.
It doesn't really have to be a prime, being relatively prime is also
good, i.e., not too many or too small factors. Surely there's a multiplier
in the right range with just two prime factors that can be computed with 3
shift-adds.
> I actually
> tried unrolling it by hand a few times, and it was slower than
> multiplication on i386 (and uncomfortably lengthy).
Right, it's not worth it unless you can get it down to a handful of
shift-adds. How does 2**17 - 1 (Mersenne prime #6) with right-shift by
(16 - bits) work?
> I believe to address architectures where multiplication is prohibitively
> expensive I should do some reading to determine a set of theoretically
> sound candidates for non-multiplicative hash functions and benchmark them.
> Knuth has some general rules about design but I would rather merely test
> some already verified by someone else and use the one that benches best
> than duplicate the various historical efforts to find good hash functions.
It would be nice if you could just look up good ones in a cookbook, but you
can't, that cookbook doesn't exist.
--
Daniel
next prev parent reply other threads:[~2002-01-05 2:41 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2002-01-04 17:40 William Lee Irwin III
2002-01-04 21:47 ` Momchil Velikov
2002-01-04 23:07 ` Andrew Morton
2002-01-04 23:21 ` Andrew Morton
2002-01-04 23:48 ` William Lee Irwin III
2002-01-05 0:37 ` Daniel Phillips
2002-01-05 1:39 ` William Lee Irwin III
2002-01-05 2:44 ` Daniel Phillips [this message]
2002-01-05 5:06 ` William Lee Irwin III
2002-01-08 18:20 ` William Lee Irwin III
2002-01-08 18:27 ` William Lee Irwin III
[not found] ` <523d1gu1ni.fsf@love-boat.topspincom.com>
2002-01-08 18:44 ` William Lee Irwin III
2002-01-08 19:20 ` Hugh Dickins
2002-01-08 20:19 ` William Lee Irwin III
2002-01-16 14:21 ` Jamie Lokier
2002-01-16 17:42 ` William Lee Irwin III
2002-01-18 0:34 ` Jamie Lokier
2002-01-06 21:09 ` Rik van Riel
2002-01-04 21:17 Ed Tomlinson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=E16Mgoj-0001Ew-00@starship.berlin \
--to=phillips@bonn-fries.net \
--cc=linux-kernel@vger.kernel.org \
--cc=wli@holomorphy.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®